Here is some of my coverage of the 2012 Silicon Valley Code Camp at Foothill College in Los Altos, California where I was reporting for Dice and Dice News.
There are three different types of dashboard design: strategic, tactical, and operational. As you move from strategic, to tactical, to operations, the level of detail in the dashboard goes up. Each serves a different purpose and a different audience, explained Lee Lukehart, Chief Data Visualist for SavvyData in his presentation “The Art of Dashboard Design” at the 2012 Silicon Valley Code Camp conference at Foothill College in Los Altos, California.
Lukehart outlined the purpose of the three types of dashboards.
Strategic dashboard: Designed for executives to monitor ongoing trends on a year-over-year basis. It’s often graphical. It’s not designed for real-time monitoring. For a dashboard to be strategic it needs a main metric you’re trying to meet, so that you can see if execution aligns with strategy.
Tactical dashboard: Built for managers to manage. Manage performance against preset target. It should be a tool to use for analysis and link results to actions. Plus the dashboard should help the manager discover problems and opportunities.
Operational dashboard: This is for the on-the-ground workers. It is where you see precise measurements in performance and other details. Information is often delivered in real time.
Avoid communications problems with effective data visualization
If not done right, dashboards can be confusing, boring, inaccurate, and worthless. Here’s how to avoid falling into that dashboard trap:
- Know when not to use a certain visualization: Pie charts are abused. They’re not efficient. In many cases, just presenting a number will be more efficient. A table or list is often more preferable.
- Know your data: Source the scope. Make sure your data is clean and complete. Avoid GIGO (garbage in garbage out).
- Consider your audience: Understand their needs, objectives, and their familiarity with the communications tools. What is their level of knowledge and experience?
- Determine the chart’s message or focus: Your dashboard is telling a story. Is it a ranking, a timeline, a category, proportion of whole, variance/deviation, distribution, correlation, or geospatial?
- Select best chart type for the message
- Construct data transforms as needed: Derive new data to tell the real story – aggregate, segment, factor, augment, find, and organize.
- Conduct pre-flight checklist: Make sure you’ve taken the following into account before you release your dashboard to the world. Consider human factors, data issues, scaling, labeling, chart type choice, visual formatting, creating an über chart, and chart titling.
Human factors in visual perception
Before you let a dashboard out the door, think about how it will be perceived. Is your dashboard creating skewed results or results certain people can’t understand?
For example, take into consideration normal human optical perception issues, such as color blindness. If you’re going to be using the color red and green to show negative and positive results, also add other indicators like maybe a plus or minus sign, or an arrow going up or down. Also, take advantage of tools within Photoshop that shows how your dashboard will look to a color blind person.
Don’t intentionally or accidentally lie with your charts. One way to misinterpret values is to not show the zero level. All charts should show the zero level so you can see appropriate relations.